Campaign Asia-Pacific teamed up with Forrester to understand how marketers and agency professionals are applying AI in their work now and plan to in the future. The final results feature over 158 senior-level respondents across Asia-Pacific, mainly from marketing and advertising agencies and brands.
The study focused on larger enterprises with over 100 employees, including multi-national agencies, non-for-profit companies and central government or public sector entities.
In a three-part series, Campaign unpacks the results of the survey and its key findings and speaks to experts on the challenges of investing in, adopting and using Gen AI.
You can read part one, which looks at investments in AI here.
Part Two: Uses of generative AI between brands and agencies
Our research with Forrester found that brands actively use generative AI, with 32% of brand respondents adopting text-prompted copy generation and 26% for image/idea generation and language translation.
Looking ahead, 82% of brand respondents plan to adopt AI for instantaneous creative optimisation, while 71% aim for real-time personalised marketing. However, over half will not use AI for text-prompted code generation (61%) and SEO workflow automation (50%) within the following year.
Meanwhile, marketing agencies are increasingly integrating generative AI into their operations, with over half currently using it for image/idea generation and copy generation.
53% of agency respondents are using text-prompted images and idea generation. For future adoption, 54% of agencies plan to use AI to summarise social and measurement data and 47% look at real-time personalised marketing.
However, nearly half have no plans to adopt AI for code generation and experiential activations in the next 12 months.
So, how are brands focusing on enhancing personalisation and content creation while showing reluctance towards specific technical applications of AI? Why do agencies strategically concentrate on content creation and data analysis while being cautious about adopting AI in certain areas? Campaign explores.
How brands are using generative AI
Singapore-based telco Singtel tells Campaign it recently integrated AI into content creation and post-production. The brand finds AI redefines storytelling and expands its creative options to enhance customer engagement.
For example, Singtel's recent 'Roam Like a Local to Any Adventure, Anywhere' campaign to promote its roaming services, Singtel ReadyRoam, involved using AI tools to generate creative assets as well as expedite the post-production process, enabling the generation of over 60 characters and backgrounds which resulted in significant cost and time savings.
A Singtel spokesperson explains the brand applied an AI filter in post-production to create comic versions of its talents.
The AI tools also enabled the production team to automate tasks like image resizing, formatting, and layout generation, usually done manually. Overall, the application of AI ensured consistency across various design materials.
"Following the successful use of AI in our campaign, we plan to continue applying AI to create more customised campaigns, which may include the incorporation of text-to-speech voice generation and the development of virtual influencers to engage our customers," says the spokesperson.
Japanese food giant Ajinomoto Frozen Food uses generative AI to engage consumers more instead of exploring conversions. How to integrate text-prompted video rendering and creation into the brand's paid media strategies to enhance customer engagement has not been discussed internally.
"We aim to increase searchability via word search, operating SEM and listing ads. We are also discussing how to manage it in the future, as AI will affect it," a spokesperson from Ajinomoto Frozen Food tells Campaign.
"The point is that the answers and insights generated by AI cannot be differentiated from our competitors, and the challenge is how we can retain our brand uniqueness."
How are agencies using generative AI?
Agencies like Publicis are using text-prompted images and idea generation as display and social advertising require quick turnarounds.
Deepa Kadam, the regional head of technology solutions for APAC at Publicis Groupe, points out that the ability to use AI for image and idea generation speeds up this process, whether it is creating quick storyboards for approval, enhancing and refining the creative idea, creating multiple versions of the master idea or generating more precise variations for personalisation at scale.
However, Deepa cautions that ensuring AI-generated content aligns with brand values and maintains a human touch is vital to preserving trust and connection with the audience.
"Thus, it is imperative to have a trained team who knows how to generate the right prompts and fine-tune the results to ensure quality," adds Deepa.
Tim Lindley, managing director of APAC at VaynerMedia, notes generative AI is another tool that furthers the agency's ability to make brands more relevant to more people at speed and scale.
He explains to Campaign that in a world where social platforms are the gatekeepers of attention, that's what drives business outcomes.
AI tools augment many of VaynerMedia's workflows. The agency strategists can contextualise data in new ways to identify and validate different audience cohorts quickly.
Creatives can scale idea generation, think as diverse personae, stress test ideas, vary the tone of voice, and so much more.
"Media teams use in-platform tools to find over-indexing creatives in real time and quickly iterate creative formats to improve performance without needing to engage additional resources or wait until formal review points. Used well, it's a major unlock to help marketers scale relevance," says Lindley.
However, Valerie Madon, chief creative officer for APAC at McCann Worldgroup, says the agency has not been able to use any output from generative AI tools wholesale, as AI cannot take into consideration many other requirements of the brief, nor does it have enough understanding to nail the brief strategically.
"Text-prompted images are only used for internal purposes, such as initial concept proposals, but are never used externally, as we are cautious of the rights of the source at McCann Worldgroup," explains Madon.
Adopting real-time personalised marketing at scale
With 47% of agency respondents in the survey looking to adopt real-time personalised marketing at scale, challenges abound in implementing these strategies across various client verticals.
There are challenges because real-time personalisation requires exceptionally detailed planning and pre-approval. Akin to the launch of Dynamic Creative Optimisation (DCO) many years ago, the planning process must be stringent, and approvals for many different variations are needed.
The shift to real-time personalisation in marketing strategies requires a combination of generative AI and human oversight. This process involves increased client involvement in approving content and establishing more nuanced audience segments for targeted personalisation.
Monitoring and measuring the outcomes of these personalisation efforts is critical to determine their effectiveness and ensure they justify the resources invested.
"On top of that, different client verticals present unique challenges. For example, in the global FMCG sector, emphasis is placed on navigating compliance hurdles and adhering to brand guidelines, while in the banking and insurance domain, the focus is on ensuring the utmost accuracy of information," explains Cheng.
"Meanwhile, within the retail sector, the challenge lies in devising sustainable workflows, considering the sheer volume of updates required. Each sector demands tailored strategies to address the intricacies of real-time personalisation."
Chris Greenough, general manager for GrowthOps Asia, stresses that effective real-time personalised marketing hinges on rapid insight generation.
He explains rapid insight generation is mainly contingent on the robustness of data collection and reporting processes and the cohesive alignment among all departments and clients in prioritising data as their guiding principle.
"Both the brand and the agency bear the responsibility to regularly refine and standardise their personalisation parameters to ensure ongoing compliance," Greenough tells Campaign.
Publicis recently created 100,000 personalised thank you video messages for employees for the agency's annual 'Wishes' campaign, demonstrating how AI can create personalised content cost-efficiently and creatively at scale.
Kadam acknowledged it was a huge challenge to pull off, and there were some glitches and inconsistencies in the output, which proves that AI is just a leverage for personalised marketing at scale and paired with human intelligence and creativity.
"It is imperative to have the right balance between automated personalisation and human creative intelligence to produce authentic content that resonates with the audience," explains Kadam.
McCann's Madon says the agency is excited by what it can achieve creatively with AI to sell and enhance brand experience and affinity.
For example, McCann created a new platform for its campaign, 'The Greatest Guide,' for its client, Bimbo, in Mexico. The platform guides users to unique Mexican street food vendors, spotlighting over 8000 chefs who innovate classic dishes with ingredients like grasshoppers and Macha sauce.
Supported by Google, it features digital maps and guides and uses generative AI for vendor branding, reflecting local art styles. The 'Greatest Guide' campaign's content drove over 77,000 visits, a high engagement rate, and increased vendor sales by 12%, contributing to a 23% category growth.
"We are very mindful that while AI is an undeniable tool for efficiency, much of its output lacks heart, and that's where our human touch is required to ensure a great consumer experience," explains Madon.
Choosing which generative AI applications to focus on
According to the survey, agencies are not prioritising text-prompted code generation, with 49% having no plans to adopt it.
While the foundation of many AI platforms relies on robust code repositories as the backbone for training prompted codes, it is prudent not to directly deploy the entire code set and instead use it primarily for troubleshooting issues.
As the technology underpinning AI prompts continues to stabilise, there should be an emphasis on evaluating the quality of the code.
"Our primary focus is on enhancing the productivity of creative and content production for performance marketing, with a parallel commitment to refining conversion rates," explains Cheng.
"Simultaneously, we are integrating generative AI into our data analysis workflows to boost productivity and facilitate rapid testing of new audience segments, thereby advancing our overall strategic objectives."
Greenough points out that most agencies are not natively digital or technical, so it is not surprising that they are reluctant to adopt technologies like Microsoft's Copilot.
"Choosing which generative AI applications to use is no different than selecting other parts of your martech stack, requiring the assessment of not only the capabilities and cost of the tool, but the ability for stakeholders to maximise its use," explains Greenough.
This story first appeared on Campaign Asia-Pacific.